Introduction

Stroke is a leading cause of long-term disability worldwide [1]. Stroke is an important risk factor for fragility fractures, due to bone loss, immobilisation, gait disturbance, vitamin D deficiency, impaired balance and increased falls risk, [2,3,4] and stroke survivors have a sevenfold increased risk of fragility fractures, which prolong post-stroke disability and increase mortality [5].

Post-stroke bone loss is focal, affecting the paretic more than the non-paretic side and is most pronounced in the first few months following stroke [3]. Despite known risks of osteoporosis and fracture following stroke, stroke patients are infrequently screened or treated for osteoporosis. A large study found dual-energy X-ray absorptiometry (DXA) testing was only performed in 5.1% of stroke survivors, and only 3.2% of those not on treatment previously were prescribed anti-osteoporosis medications within 12 months [6].

In recent years, fracture risk scores have been developed to assist in fracture prediction following stroke [7, 8]. The Fracture Risk after Ischemic Stroke (FRAC- Stroke) score was derived from the Ontario Stroke Registry and incorporates patient age, female sex, modified Rankin Scale (mRS), rheumatoid arthritis, osteoporosis, and previous falls and fracture. A FRAC-Stroke score threshold is recommended to identify stroke patients who should be screened and treated for osteoporosis: patients with a 1-year fracture risk of greater than 0.8% (FRAC-Stroke Score ≥ -3) are recommended to be screened with DXA while those at very high fracture risk (1-year risk, ≥ 2.0%; FRAC-Stroke Score ≥ 6) should be considered for empirical osteoporosis therapy regardless of bone mineral density (BMD) [8]. A recent Australian registry study also found common factors such as age, female sex, history of previous fracture associated with increased risk of fracture post stroke [9]. There are, however, no current recommendations for testing or treatment of osteoporosis in stroke guidelines. In addition, very few clinical trials [10, 11] address drug treatment for the prevention and treatment of osteoporosis and fracture after stroke.

This study aimed to study the prevalence of osteoporosis, falls and fractures in adults with ischaemic stroke. Using this data, the utility of a bedside fracture risk assessment tool, the FRAC-Stroke score, will be assessed in an Australian cohort. Validation of this score has not been performed outside of the Ontario population. If applicable, this will allow rapid identification of high risk patients to be referred for management of bone health.

Methodology

This was a retrospective pilot cohort study of adults aged ≥ 50 years admitted with ischaemic stroke at Monash Health from 1 January 2020 to 31 December 2020. Inclusion criteria required admission under the stroke team and imaging confirmation of ischaemic stroke either with computed tomography (CT) or magnetic resonance imaging (MRI) of the brain. Patients were excluded if they had haemorrhagic or other types of non-ischaemic stroke, had no mRS documented, or had passed away by the time of recruitment for telephone interviews. Demographic data, medical history, relevant biochemistry and 90 day mRS score was obtained from electronic medical records to assist with calculation of the Fracture Risk After Ischaemic Stroke (FRAC-stroke) score. Modified Rankin Scale measures the degree of disability following stroke and ranges from 0 with no disability to 6 where 6 is death. For example, an mRS of 5 is severe disability, bedbound, incontinent requiring nursing care. All patients identified were invited to participate in telephone interviews to ascertain falls and fracture 12 months post-stroke as well as confirm medical history obtained from the medical records. This included questions regarding smoking history, alcohol intake, use of steroid medication and diagnosis of rheumatoid arthritis or other inflammatory arthritis (see Supplementary Material).

The FRAC-stroke score can range from -9 to a maximum of 36 and is calculated on 6 domains of age, sex, mRS, rheumatoid arthritis, osteoporosis and previous falls or fracture. For example, in the age domain, age 80–89 years scores highest with a score of 8 whereas age 50–59 years scores lowest at -3. Female gender scores 5 points whilst male 0 points. Previous fracture scores highly with an addition of 10 points and previous falls scores 3 points.

Minimal trauma fracture was defined as a self-reported or radiologically proven fracture occurring after a fall from standing height or less, or a minimal trauma incident other than a fall (e.g. turning over in bed). Smoking was defined as current smoker and alcohol use was 3 or more standard drinks a day consistent with the Fracture Risk Assessment tool (FRAX®). Similarly, consistent with FRAX definitions, glucocorticoid use was considered significant if oral glucocorticoids were used for 3 months or more at a dose of at least 5mg prednisolone or equivalent. The study was approved by the Monash Health Research Office.

Statistical analysis

The distribution of the data was explored by the Shapiro–Wilk test. All normally distributed data was expressed as mean with standard deviation and non-parametric data as median with interquartile ranges. Subjects were divided into 2 groups, those who had fractured in the 12 months post-stroke and those who had not. Differences between groups were determined using the independent t-test for continuous variables and chi square for categorical variables. Group differences in mRS were determined using the Kruskal Wallis test. Area under the receiver operator curve analysis (ROC) was performed to determine the predictive value of the FRAC-stroke score for fracture in the 12 months following stroke.

The optimal cutoff was determined using the maximum value of Youden’s index (J = sensitivity + specificity-1). Analyses were conducted using IBM SPSS statistics for Mac (Version 29.0.1 Armonk, NY).

Results

1267 patients were admitted to the stroke unit between 1 January 2020 and 31 December 2020 with presumed ischaemic or haemorrhagic stroke. Of these, 624 patients had a mRS score documented and 316 adults ≥ 50 years were identified with ischaemic stroke (Fig. 1). 131 of 316 adults with ischaemic stroke consented to a telephone interview (see Supplementary Material). Baseline characteristics are shown in Table 1. Mean age was 72.4 ± 10.7 years and 36.6% were female. Renal impairment was common with 29% having chronic kidney disease (CKD) stage 3A or higher. Current smoking was highly prevalent at 19.1%. Vitamin D levels were only available in 24 patients within 6 months of stroke.

Fig. 1
figure 1

Methodology of study

Full size image
Table 1 Participant characteristics
Full size table

Fourteen (10.7%) had a MTF prior to the stroke (12 female) and 17 (13.0%) had a history of osteoporosis. Only 10 patients were receiving treatment for osteoporosis and 14 (10.7%) had a previous DXA scan. There was a significant difference in patients experiencing falls pre- and post-stroke (19.8% vs 31.3%, p = 0.04). 38.5% of patients who had a fall pre-stroke experienced a fall post-stroke.

Eighteen patients (13.7%) had a FRAC-stroke score of between 15–22 equating to a 5–10% risk of fracture in the year following stroke. Sixteen (12.2%) had a FRAC-stroke score of ≥ 23 equating to a > 10% risk of fracture in the year following stroke. Eleven (8.4%) patients (6 female) had a MTF in the 12 months post-stroke; six fractures occurred in the upper limb and five in the lower limb/pelvis. Patients who fractured post stroke were older, and more likely to have a history of falls, osteoporosis or fracture prior to the stroke (see Table 2). FRAC-stroke score was higher in those who had a fracture post stroke compared those who did not (20.4 vs 8.9, p < 0.001).

Table 2 Comparison of patients who fractured 12 months post stroke to those without fracture
Full size table

Considering previous cut-offs suggested, all patients in our cohort would require DXA screening (FRAC-stroke score ≥ -3) and 64.1% would require empirical treatment (FRAC-Stroke score ≥ 6).

ROC analysis found an AUC of 0.867 for the FRAC-stroke score (95% CI 0.785–0.949, p < 0.005) (Fig. 2). According to Youden’s Index, the optimal cutoff value for FRAC-stroke score predicting fracture was 12 with a sensitivity of 90.9% and specificity of 70%. 38.9% of our cohort had a FRAC-stroke score of ≥ 12, and of these patients 19.6% had a fracture in the 12 months following stroke.

Fig. 2
figure 2

ROC curve analysis for FRAC-stroke score prediction of fracture in 12 months following stroke

Full size image

Discussion

Stroke has long been known as a risk factor for osteoporosis and fracture. 8.4% of our patients experienced an MTF in the 12 months following a stroke. This cumulative incidence rate is comparable to that of the 10% rate of subsequent fracture in the 12 months following initial clinical fracture in postmenopausal women [12]. Stroke should therefore also be considered a risk factor for imminent fracture and timely management is critical. This rate is higher than that reported in an Australian registry study [9] which found 3.8% of patients experienced a fracture within 1 year of stroke or transient ischaemic attack (TIA), however they included adults ≥ 18 years of age and all types of stroke and patients who had TIA.

We found that those who fractured following stroke were older, and more likely to have had a history of falls, minimal trauma fracture or a diagnosis of osteoporosis prior to their stroke. Similarly, Forster et al. and Dalli et al. have also found these risk factors for fracture following stroke as well as female sex and malnutrition [9, 13].

To our knowledge, we are the first study to validate the use of the FRAC-stroke score outside of the Ontario population which was used to derive and validate the score initially. As a clinical tool, the FRAC-stroke score can be calculated easily by the bedside prior to discharge from a stroke unit. Smith et al. originally proposed a cut off point of FRAC-stroke score ≥ -3 for screening and ≥ 6 for empirical treatment [8]. These cut offs would be overly sensitive where all patients in our cohort would require screening and over half would need empirical therapy. As such, we propose a cut off score of ≥ 12 for empirical treatment, based on our ROC analysis. A single dose of zoledronic acid has previously been shown to prevent bone loss from the hemiplegic hip after acute stroke and may be an attractive intervention [11].

The mechanism by which stroke leads to fracture is multifactorial. We found a significantly increased risk of falls post-stroke with rates of 31.3% post-stroke compared to 19.8% pre-stroke. Impaired functional mobility, neglect, polypharmacy, cerebellar and visual field defects may all contribute to falls [14]. Falls prevention will be key in reducing fracture rates, and a recent Cochrane review demonstrated that exercise may reduce rates of falls in the post stroke setting [15].

Limitations of our study include lack of vitamin D levels and BMD given its retrospective nature. We did not have radiology to confirm fractures which were self-reported. Our sample size is small and our findings need to be replicated in a larger cohort. The role of the FRAC-stroke score in haemorrhagic stroke or TIA is yet to be elucidated and cannot be generalised to a younger population. Our proposed cutoff for empirical treatment needs to be validated in another cohort to avoid overtreatment.

In conclusion, FRAC-stroke score is easily calculated by the bedside, with a score of ≥ 12 predicting those at high imminent risk of fracture. Empirical treatment should be instituted to prevent further functional limitations and disability in stroke patients.